Currently, if a user uses the --launch-command option --to launch the job on a compute node-- and something goes wrong (e.g. they specified a non-existing queue or something is off in their job submission), jupyter-forward gets stuck in a re-try loop. We should make sure to detect this kind of failures and report them and exit.
Currently, if a user uses the
--launch-commandoption --to launch the job on a compute node-- and something goes wrong (e.g. they specified a non-existing queue or something is off in their job submission),jupyter-forwardgets stuck in are-tryloop. We should make sure to detect this kind of failures and report them and exit.